445nm specific PBS cube GB

[buffo]

Even the LM317 data sheet shows that you can gang them.

Yeah, I'm not really comfortable with that. How do you ensure current is evenly shared between them? (Given that if one shorts, your diodes go poof!)

And with the LM338, directly compatible in this case, you might not even need to do that.

Now that's an idea... It can handle 5 amps (with proper heatsinking), so that might work.

As to risk, try it on something less important

True. I actually built a large stack of 4008 diodes for exactly this purpose after my latest Flexmod failure...

Hmmm. Ok, you've got me thinking about it at least!

Adam

[The_Doctor]

Nice. Go for it. Re ganging LM317 or LM338, you use one regulator, and one resistor to hard-limit the current. Each diode gets one set like that, and each diode/regulator/resistor set gets controlled by a single driver! So long as the ratio of threshold current to full current is the same for each diode driven, you could mix WILDLY different diodes on one driver. Not much point but that illustrates how versatile it is.

Given the voltages possible you can even chain them in series, but I'd rather use additional regulators. Main thing then is juggling costs of big heatsinks versus costs of diodes. If diodes are cheaper, series might be good...

EDIT:
Do you really mean 4008 diodes?! Last I looked they went up to 4007 (1000V type) for standard 1A rectifiers. If they did extend to 4008 I ought to play with some.

[buffo]

Nice. Go for it. Re ganging LM317 or LM338, you use one regulator, and one resistor to hard-limit the current. Each diode gets one set like that, and each diode/regulator/resistor set gets controlled by a single driver!

I'm not sure we're expressing the same setup. I'm talking about driving one diode at high current. So you only have one diode, but you still need multiple 317's to get the higher current. (Though admittedly, a single 338 would work up to 5 amps.)

Are you saying you would gang together all the Vadj pins from all the 317's and connect them to the output of one driver (the output of U1B)? So each 317 would have it's own R11? And you're relying on that resistance to provide current feedback as well as over-current protection for each regulator?

Do you really mean 4008 diodes?! Last I looked they went up to 4007 (1000V type) for standard 1A rectifiers. If they did extend to 4008 I ought to play with some.

No - I derped on the keypad. I meant 4002. Standard 1 amp rectifier diodes.

Adam

[The_Doctor]

Are you saying you would gang together all the Vadj pins from all the 317's and connect them to the output of one driver (the output of U1B)? So each 317 would have it's own R11? And you're relying on that resistance to provide current feedback as well as over-current protection for each regulator?

If the regulators were from same batch, yes. Best to test them for sufficiently close behaviour, and definitely mount them on the same heatsink. With each resistor being 1% tolerance metal film it will be ok. If a laser diode is pushed to the point where 1% makes a fatal difference, it's definitely being pushed too hard.

EDIT:
In the LM317 datasheet it shows an application, "Adjustable 4A Regulator" that shows how Adj and In are all commoned while Out gets its own resistor, and just one of those is watched. That laser driver circuit is controlling current not voltage, but this is within design intentions for that regulator. It will work the same way.

More edit:
In a case where more than one diode were driven (each with its own regulator and sense resistor), as well as watching that all regulators were from same batch and mounted on one big heatsink, it pays to check that each diode has same Vf! Failing that, I'd use a driver for each, or consider running diodes in series if I knew their threshold and max currents were the same. In this last case, difference in Vf amount purely to differences in heat dissipation, but good thermal clamping per diode takes care of that. The main risk with this scheme is that good thermal clamping isn't always easy to acheive if you have to electrically isolate diode cans from each other as needed there.

[buffo]

Back to the PBS topic for a bit:

I got my diodes up and running today, and managed to run some preliminary tests on the PBS cube that Dave sent me. I am seeing roughly 10% loss through the cube, on both the horizontal and vertical. That seems to be livable... EDIT: After letting everything stabilize thermally, I'm actually seeing more like 5%, which is much better.

Angle of incidence is *not* a major factor, which is really nice. I have a pair of TEC-cooled M140 diodes running at 1.6 amps CW right now, and they're baking the cube at point-blank range. I'm getting around 3.3 watts out the front. I'll let it cook for a day or so and see if there are any changes in the faces, or if the losses increase.

I also have a larger PBS cube I bought from Mike Lawson (Mecheng3) many years ago, so I decided to test it against this one from Dave. By comparison, the one from Mike has similar losses, but it is very sensitive to the angle that the beam enters the cube. Just a few degrees difference causes the losses to shoot up to 30%! So far I would much rather work with the cube from Dave, even though it's rather small. (Still more than large enough for the beams from a dual 445 rig to fit.)

I'll post more about this test tomorrow, but for now here is a picture of the burn-in (click for full-size version):



Dave's cube is on the right. The larger cube to the left (and further to the rear) is the older one I got from Mike. The vertically-polarized beam is shooting straight through the cube and striking the metal face of the power meter in the background. (Not the sensor, just the aluminum face plate, which is why there is so much reflected 445.)

Note that I'm only testing the vertical polarization here. I need to get my other set of diodes up and running in order to do the full scale test with all 4 diodes. That will be later...

In the LM317 datasheet it shows an application, "Adjustable 4A Regulator" that shows how Adj and In are all commoned while Out gets its own resistor, and just one of those is watched.

Ahhh... That's the part I missed. Each regulator has it's own resistor on the output, but only one is connected to the input of U1B. Got it.

OK, I can see how that would work. I think I need to try this.

In a case where more than one diode were driven (each with its own regulator and sense resistor), as well as watching that all regulators were from same batch and mounted on one big heatsink, it pays to check that each diode has same Vf! Failing that, I'd use a driver for each, or consider running diodes in series if I knew their threshold and max currents were the same.

Yeah, I'm running my 445s in series. Admittedly though, the 445s are pretty hardy, so as long as you aren't right up against the hard limit, a little variation probably wouldn't matter.

Adam

[The_Doctor]

That inspires me. Small PBS not sensitive to angles. I like it. 10% sounds like a lot but maybe my expectations are wrong, this isn't like HR mirrors, and those would be sensitive to angles too.. Really nice mix of colour in that image. What lenses are you using on those diodes? I have some Aixiz ones with nonstandard threads I never got round to trying. I'm wondering if I should, or get something else instead. I don't mind losing a lot of light with these, as there's so much of it! I want to get a nice narrow beam of bright well collimated blue with about 100 mW or so in it. I guess that what the Aixiz lenses are after doing anyway..


Interesting that you run in series. I always saw no great risk there, but many here railed against the idea. But as I say, it neatly eliminates concerns over differences in diode Vf when running two off one driver.

Btw, The one blue diode I tested so far, I borked, big time. And before it augured in it did something weird, reduced output due to overheat in a way that was (initially) recoverable. As far as I know, my heatsink, though inadequate, wasn't a lot more than blood-hot at the time! Are those diodes especially prone to weirdness with temperature change? I thought they were designed to run hot, in those projectors. I never figured that out, and held off serious tests in hope of more info that I never ended up searching for very hard.

[buffo]

That inspires me. Small PBS not sensitive to angles. I like it.

Yeah, so far they have been relatively insensitive even to widely different angles of incidence. My other, larger cube is nothing like this.

10% sounds like a lot but maybe my expectations are wrong

See my edit above. I was powering them on and off, and with the TEC running flat out, the diodes were getting quite cold when the power was off. So the first power measurement (no cube) ended up being on the hot side. Then, by the time I got the cube in place and repeated the measurement, power had drifted down some as the mounts heated up to normal temp.

I've had the thing running steady state for over 4 hours now, and when I just checked I had 3.24 watts coming out of the diodes and 3.07 watts coming out of the PBS. So that's ~ 5 % loss, which is much better. I'm going to let it bake for a day or two like that though.

What lenses are you using on those diodes?

The G2. Seems to work really well with secondary correction (prisms or cylindricals). Some people swear by the O-like lens, but I've always had good luck with the G2, so I've stuck with it despite the increased cost.

I want to get a nice narrow beam of bright well collimated blue with about 100 mW or so in it.

Then you might be better off buying one of the single-mode 445 diodes. I'm pretty sure they put out well north of 100 mw, and the beam is excellent right out of the package. (No secondary correction needed.)

Interesting that you run in series. I always saw no great risk there, but many here railed against the idea. But as I say, it neatly eliminates concerns over differences in diode Vf when running two off one driver.

Mostly it's to lessen the heat load on the driver. If you have to run off 12 V, you're better off powering a stack of 2 diodes in series, since that way the output transistor is only holding back a volt or two. But yeah, the risk to the diodes is trivial.

Parallel operation, on the other hand, is far more risky.

Are those diodes especially prone to weirdness with temperature change?

In my experience, yes. They like to stay cool, and will produce more light when they're kept cool. But even when they get really hot, they will still lase (to a point, of course), although the efficiency drops way off when they get hot.

They really are pretty tough, and many people run them hot with nothing but passive cooling. Nevertheless, I've had more than a few of them taken out by driver failures, so they're not bullet-proof.

I thought they were designed to run hot, in those projectors.

In the original Ca$io projectors, the diodes were mounted in a large aluminum block. It wasn't a bad heat sink, apart from the shear number of diodes (24) adding heat to it. Still, it worked reasonably well. The diodes got hot, but not scorchingly so.

I've seen A140 diodes (the original 1 watt versions) in an Aixiz module clamped into one of Pat Bischoff's mini-mounts and running right at 1 watt continuously for hours. I promise you that diode was hot as hell, but it never died...

Here's a picture of the mount I'm talking about: Notice how little surface area there is to conduct heat away? And yet the diode lived. So yeah, they're pretty durable.


A better solution is to use a Z-bolt heat sink, but that's still passive. I'm using the brass cube mounts from Dave, and I have a large TEC under that plate to cool everything down. I think a TEC is the idea solution for these things, but I do admit that it's a luxury, not a necessity.

Adam

[The_Doctor]

I've had the thing running steady state for over 4 hours now, and when I just checked I had 3.24 watts coming out of the diodes and 3.07 watts coming out of the PBS. So that's ~ 5 % loss, which is much better. I'm going to let it bake for a day or two like that though.

Very nice. I begin to wonder how many more reefs made of large frame argons may grow mysteriously off the shores of Florida and like places.

The G2. Seems to work really well with secondary correction (prisms or cylindricals). Some people swear by the O-like lens, but I've always had good luck with the G2, so I've stuck with it despite the increased cost.

If you're able to get anything like a 2mm by 2mm beam with less than 0.5 milliradian divergence out of that with anamorphic prisms I'm totally convinced.

Then you might be better off buying one of the single-mode 445 diodes. I'm pretty sure they put out well north of 100 mw, and the beam is excellent right out of the package. (No secondary correction needed.)

That really DOES convince me! This is what a few years away from PL do. I had no idea that single mode high power blue was in affordable reach. Any idea how much? This might become priority number one, especially as I read that single mode green is also beginning to appear. I love the reduction of complications by using TEM00 beams.

They like to stay cool, and will produce more light when they're kept cool. But even when they get really hot, they will still lase (to a point, of course), although the efficiency drops way off when they get hot.

They really are pretty tough, and many people run them hot with nothing but passive cooling. Nevertheless, I've had more than a few of them taken out by driver failures, so they're not bullet-proof.

That fits with my observations. Wasn't a drive failure in mine, I now remember how it died... I tried retroreflection on it. Ever since I discovered this horrible death for a diode at times when people were telling me it couldn't happen, I have had the urge to find out for sure the only way to know. I guess that all diodes will die that way unless below maybe 50 mW or so. I guess it's because it puts several watts on both sides of the facet by extending the cavity. Fast and brutal... A bit like one of my favourite bands, Pissing Razors, but I digress...

In the original Ca$io projectors, the diodes were mounted in a large aluminum block. It wasn't a bad heat sink, apart from the shear number of diodes (24) adding heat to it. Still, it worked reasonably well. The diodes got hot, but not scorchingly so.

I've seen A140 diodes (the original 1 watt versions) in an Aixiz module clamped into one of Pat Bischoff's mini-mounts and running right at 1 watt continuously for hours. I promise you that diode was hot as hell, but it never died...

I let mine get very hot. Actually even retroreflection never totally killed it, there was still a bit of specular activity in it afterwards. And even if I let it get hot, a tad bit of that still remained. Light spill was disgusting though. I don't mind losses, given all that power, but I don't want to have to SEE them.

Here's a picture of the mount I'm talking about: Notice how little surface area there is to conduct heat away? And yet the diode lived. So yeah, they're pretty durable.

That's tidy. The proportions of my little brass blocks soldered with indium-based solder might do it a bit better, though brass won't conduct as well as Al. I like brass though, solderable, and it makes very tough threads using the tiny opticians screws that Jem got for me. It will be great to try working with that stuff again.

A better solution is to use a Z-bolt heat sink, but that's still passive. I'm using the brass cube mounts from Dave, and I have a large TEC under that plate to cool everything down. I think a TEC is the idea solution for these things, but I do admit that it's a luxury, not a necessity.

A TEC is a good idea. The way I see it, if you can drop the system temperature while raising the case temperature, the heat difference from ambient helps it get rid of heat. If the TEC can stand the hot side temperature a system could run cool in a very hot location. If this can favourably shift wavelength as it does for reds, so much the better..

[buffo]

If you're able to get anything like a 2mm by 2mm beam with less than 0.5 milliradian divergence out of that with anamorphic prisms I'm totally convinced.

I've never gotten results that good. More like 2mm x 4mm at the aperture with roughly 1 mrad divergence. But others may have had better luck.

Now, the single-mode diodes are an entirely different matter. I haven't actually seen the beam from one in person, but I've been told that they're worlds better than the high-power 445 diodes. They might actually hit your specs of 2mm square and less than 1 mrad...

I had no idea that single mode high power blue was in affordable reach.

Well, the single mode blues aren't really "high power". They top out at something like 120 mw. By comparison, you can get nearly 3 watts out of a single 445 nm diode in a 9mm can, but it's multi-mode and will definitely need secondary optics to correct the fast axis.

The single-mode units sell for under $50, and the high-power multi-mode diodes are between $50 and $90, depending on which one you get. (A140, M140, or the 9mm can units)

A TEC is a good idea. <snip> If this can favourably shift wavelength as it does for reds, so much the better..

I believe the wavelength shift per degree C of temperature change is very small, like .1 nm or something, so unlike the reds, cooling doesn't really change the color very much. (And in any case, cooling would tend to make the wavelength shorter, which is the wrong way to go if you're starting at 445. We want a blue that's closer to 460 nm, not 430...)

Adam

[dave]

I've had the thing running steady state for over 4 hours now, and when I just checked I had 3.24 watts coming out of the diodes and 3.07 watts coming out of the PBS. So that's ~ 5 % loss, which is much better. I'm going to let it bake for a day or two like that though.

Nice one

One thing tho - I was testing much lower losses than 5% with that particular cube - you have rotated diodes for max thruput?